
Loading summary
A
For as long as Azure has been around, which is 15 years as of this year, Microsoft and AWS have been competing. Sure, AWS is bigger and has been at cloud computing longer, but the Azure team has continued nipping at their heels, announcing price cuts and new services which AWS consistently counter announces against. But now we're in the AI era, which is different. I think Microsoft has more mindshare maybe and market share possibly. And but Azure and AWS are vying to sign up all the right AI darlings as partners. And now Microsoft has started deploying anthropic models, which AWS was doing already. And AWS recently announced a multi billion dollar, multi year partnership with OpenAI, Microsoft's key AI partner. So my question is there any competitive advantage for one of these hyperscalers over the other anymore? Or. Or are we in a world where AI has commoditized the cloud? Welcome to the Directions on Microsoft Briefing Podcast. I am Mary Jo Foley, the editor in chief here at Directions. I am your host for the series of podcasts for those interested in the Microsoft enterprise IT ecosystem. We have two great guests today to discuss Azure AWS commoditization. Corey Quinn is here today. He's the chief cloud economist at Duckbill, where he specializes in helping companies improve their AWS bills by making them smaller and less horrifying, as he likes to say. And we've got Directions analyst Wes Miller, who covers Microsoft security identity management and systems management technologies. He previously worked at Microsoft as a program manager in the Windows Core operating system division and for Winternal Software, which was acquired by Microsoft in 2006. Hi guys. Thank you for being on the podcast today.
B
Thank you for having us. It's always a pleasure for me to indulge my ongoing love affair with the sound of my own voice.
C
Indeed. Thank you very much, Mary Jo.
A
All right, Corey, we're going to put you on the hot seat just right out of the gate. We know you just got back from aws re Invent. Wes and I attended Ignite, but virtually both shows, of course, were heavily focused on AI agents and bears. Oh my. Do either of you think there's still a substantial difference in how AWS and Microsoft are approaching, approaching the AI space right now? Or are they basically doing the same thing? What do you say, Corey?
B
Their approach to it has been largely, yes, they are launching their own models, they are signing deals with the other folks, and we see the financial ouroboros as Nvidia starts doing some things that look not too dissimilar from what MCI World Comp did once upon a time of self financing partners and customers, they could afford to spend more with them. And 2026 to 2027 is going to be really interesting in that space. But as far as capabilities go, if I had to pick a provider that had differentiated capabilities out of the hyperscalers, I'd say Google these days.
C
Interesting.
A
All that thought, Wes?
C
Oh, I just think it's. I agree, I agree with Corey. I think it's interesting. Interesting space. Everybody sort of evened out at this point. So it's. We'll see where they go from here.
A
Yeah, it really does feel like that. Like even today there was a big announcement about an agentic AI foundation and I'm like, yeah, AWS is there, Google's there, Microsoft's there. What's the difference, right? Everybody's there. Okay, so what do you guys think of my initial hypothesis I started out with? Which is, are the two cloud platforms basically commoditized now because of AI or in spite of AI or are there still some actual distinct reasons a customer should go with AWS or Azure? Corey, you want to start?
B
Oh, I'm not going to win friends on this. Generally between 5 to 7% of company infrastructure spend as a ballpark direction among my customer base is where I spend is going. That still leaves a massive majority of things that are. How do I put this? Not ridiculous. When it comes to run the care and feeding of their ongoing business needs, Azure remains the less expensive option generally because their historical security has been so weak that you can just run stuff in other people's accounts and it's hard to cost optimize beyond that. Whereas AWS has sort of been competing heavily with Azure in both of them. In the space of what? How can we give something a worse name for a product? Azure DevOps was a close contender. Systems manager, session manager came in clutch and honestly everyone is just confused now they're both saying the same things. But if you remove the names and any of the identifying specifics, you can't pick up the difference between one to the other in terms of feature approaches, infrastructure reliability. Yes, there are differences. Ability to put workloads where customers want, obvious differences. Ability to reach out as an annoying customer and get attention from these providers. There's some differentiation there. But my advice remains the same as it has been since I started. Pick a horse, go all in. I don't particularly care which one you pick. I'm not a partisan here, Wes.
C
Yeah, I definitely think we're still at the point where most companies are Spending so much on things that aren't AI. You know, I hate the term, but the table stakes that they've been spending on for the last 15 years in Azure, and that's the same in Azure, Google, aws, and I think we're still at this early stage where companies are dabbling in AI, so the amount they're spending is small. But to me, it doesn't make sense to pick a partner for AI that's not your cloud partner. If you're already betting big on Azure, keep going with Azure. If you're already invested deeply in one of the other two, go with AWS or go with Google Cloud.
A
Okay, well, Corey broached the G word, Google, and I have a question there. I always feel like industry watchers are quick to dismiss Google in the cloud wars. They're always like Google, a distant number three, right? But I wonder if AI has changed this dynamic. I noticed at Re Invent aws and Google announced a big networking partnership. I don't know if it was just for show or it has teeth, but what do you guys think of Google these days? Are they more of a competitor because of AI? Who wants to start on that one?
B
Wes, why don't you take this one to start? I've started the last few, and otherwise it sounds like I'm just backing you into a corner here.
C
It's okay, it's okay. I appreciate it. I think. I think when you look at Google, to me, we always approach everything, you know, I see it from a different perspective, as Corey does, because of what our businesses each focus on. And so we innately wind up talking to customers. AWS is sort of the second thing we talk about. Even though a lot of them are much more deeply invested in aws, Microsoft is, let's just say, encouraging them to look at Azure for more and more investments over time. But Google, I think, still keeps coming up. And to me, I keep hearing Google, you know, at the platform layer. I don't hear Google a lot still. I think they're getting mind share in the AI platform space. But to me, things like Gemini, you know, at the higher level, are getting more mind share than, for example, Microsoft's, you know, cavalcade of copilots which has just been so confusing. Customers don't know what anything's called anymore. So I think. I think we're. We're almost at a point where Google may have a different advantage than the other two. And so we'll have to sort of see where that plays out over the next year or two.
B
Wholeheartedly agree on that. But one nuance I would put in there is everyone is in bed with Microsoft whether they want to be or not for Office365 purposes, for Windows purposes. Every company of any size has that has agreements there. And how does that parlay into Azure adoption? Even my customers don't really know how much they're spending on Azure versus the rest of Microsoft. It's extremely unclear when you approach from a pure cloud infrastructure perspective for workloads that are Internet facing. I see GCP more than I see Azure and I think that's going to continue to be a large sweeping Trend, even putting AI to the side. I don't know what AI is going to look like in 10 years, but I promise you we are still going to be running databases badly. We're still going to need web servers that serve web there's compute, storage, networking. Those things are not going away. Trying to say oh that's legacy stuff maybe, but mainframes are still here however many decades later too.
A
Very true, very true. All right, I'm going to take this opportunity to take a break so I can talk about why you should make it a priority in 2026 to go to one of the directions on Microsoft licensing and negotiation boot camps. And not just because Wes is going to be there. That's a good reason, but not the only reason. As of November 1, 2025, Microsoft eliminated tiered volume discounts for online services in the EA and Microsoft Customer Agreement or mca. So if you're not prepared for this, this change means you're going to face a potential 6 to 12% price increase just on your core Microsoft cloud services like Microsoft 365, Azure and Dynamics. So what's your best defense? You need to invest in your Microsoft licensing expertise. If you go to our next in person directions on Microsoft Licensing and contract negotiation bootcamp, it's going to be in Los Angeles February 3rd 5th, 2026. You should send your IT procurement and ITAM leaders there to learn how to deal with these new Microsoft licensing realities. You'll learn things like how to control your Azure consumption costs so you can cut your largest Microsoft cloud spend. We'll help you master the strict licensing rules for high cost products like Copilot and Microsoft 365 add ons. We'll help you figure out how not to over buy while maximizing your roi. If you want to get your spot secured today, go to directions on Microsoft.com training. You'll get to talk to Dean Wes and our other directions for licensing experts there. And if you can't make la, we are going to do boot camps in person in Washington D.C. and Chicago next year. So make sure you join us. Corey, you should join us.
B
I'm sorely tempted. The thing is is that I can do Google negotiation and AWS negotiation and I do a lot when it comes to the Microsoft stuff, I tap out because I'm not an auditor, I'm not an attorney and I know you folks. It is easier to tag you folks in. So much of it is down to license, interpretation and auditability. I used to be a Windows admin once upon a time and I got out of it because I couldn't stand the license audit Gotcha stories. It drives me nuts. I'm not constitutionally suited for it and I know folks, why would I possibly want to do it myself?
C
We feel your pain.
A
You can still come and hold our hands. That would be good.
B
Yes, I will wrap whatever it is you're talking about. Why not? Everyone loves that.
A
Nice. All right, let's go back to the Core and West show all about Azure and aws. So when we're talking about competitive advantage, one place that Microsoft seems very focused, understandably, is the enterprise beachhead it has with Copilot. So I'm curious, do you guys think this gives them an advantage or a disadvantage or is it just impossible to tell with their underlying AI platform on Azure? Anyone have thoughts there? Wes?
C
Corey, I think, I think my, my personal opinion is that it's not an advantage. You would think it, you might think it would be, but I think it's a lot to do with the confusing branding and the fact that the branding within the platform space and the branding within Microsoft 365 really don't feed off of each other. There's no virtuous cycle getting organizations to say, you know what, this is built on that great platform in Azure. We should be using that too. So I think they're sort of missing the boat with that opportunity and it'll be interesting to see how it plays out. But I don't think that's going to change until they realize that there's sort of that loop that they need to close.
B
Yeah, the branding is a complete mess. I mean, every company is doing this right now. It seems everything is Gemini over on the Google side and everything is Amazon Q over an but on the Microsoft side it is so invasive, which tells me that it's a push motion rather than a pull, which says a lot. I mean it's effectively what if Microsoft Clippy came back, only now he's been reading your mail. It's not a great customer experience historically compared to the other models. It's always felt a little bit more paternalistic. Rather than let me help you with the thing that you're doing. It takes an approach of clearly you don't know how to write, let me do it for you. Which personally I find insulting.
A
So do we as right.
C
Yeah, you're not alone, Corey. I saw a similar piece of feedback on Bluesky. Someone basically saying the exact same thing over the weekend. They kept a thread of 10 things that Clippy did that were annoying over the weekend, which is not a great thing to see.
B
Yeah, it's. It's just insulting because it's There are things that there are things that computers can do that we do, and there are things that we do that computers have not yet figured out the way to do this. So the question you really have here is what are the differentiators between you and AI both today and where it's going? Well, there are some, there are some answers to that, but none of them involve and then I've abdicated that whole pesky thinking thing to the computer.
A
Right. I think the tone thing too. Right. Like that Microsoft struggled with this and they all do. But even back to the old Cortana. And before that, Bob.
B
Right.
A
Like trying to make it not sound condescending when they're trying to quote, help. That's tough.
B
Oh my God, yes. It's like a Google employee came to life.
A
All right, there was an interesting interview at Re Invent, the acquired podcast team did with AWS's CEO Matt Garman. By the way, if you don't know about the acquired podcast, those guys, they know how to get people talking. I got interviewed by them and I said way too much when I was talking to them, but it was fun. So Matt Garmin said a couple interesting things during this podcast. First, he said Bedrock is a multi billion dollar business already. So when I think of Bedrock, I think of it roughly as the equivalent of what Microsoft now calls Microsoft Foundry. You should be Azure. AI foundry probably had 10 other names before that. And it's where developers can choose from models and develop apps and agents. So the information reported recently, Microsoft struggling to get customers to see value of Foundry. So I wonder which is it? Are agent factories, as Microsoft call them, a real thing? Are they an aspirational thing or not even a thing? What do you think? Corey?
B
I can guess. I do know that by and large I see customer adoption of Bedrock being fairly high. I have one particular customer that I talk to though that spends more on Anthropic Direct directly than they do on aws. And for their business it makes sense, but that is very much an outlier. I want to be clear on this. I use it myself for a bunch of weird things, but I don't see that right now saying that, oh, it's a multi billion dollar business. Great. Aws is a $132 billion business annually run rate right now.
C
So.
B
Okay, great. It's one like the service we talk about exhaustingly and does have a lot of value, is now over 1 to 2% of our annual revenue. It's good for you. You're beating out of a long tail of services that are not that AWS config was trying hard but couldn't make it. I mean, what's the point there?
A
Yeah. Yep. Wes, thoughts?
C
I totally agree. I think it's a. They're both trying to get the. They're slowly beating the drum louder and louder in order to try and get attention for their platform. And I think, you know, it's, it's, again, it's so small in the grand scheme of things. When you look at what the rest of the platform does and really what the rest of customers are doing with that platform that it's, it's, it's odd that we wind up having to spend so much time looking at it and watching it and analyzing it closely given the space it's in right now.
A
It's true.
B
Yeah. Tea leaves here. And for all we know, because he was speaking off the cuff and depending on phrasing, he could have just said it's a multi billion dollar business to run. I mean, babbling about it every chance you get in every corner. I mean, yeah, that, that takes serious marketing dollars.
A
Does it? Does. Yeah. I don't know. I really feel like when Microsoft stopped talking on their earnings call about what percentage of our Azure revenue is actually attributable. Attributable to AI. Whenever they stop giving those numbers, there's a reason. Right. And I think they were like, yeah, we probably shouldn't tell people what that percentage is. Right.
B
Well, I also don't trust any number you'll give me. It's the. You take this ide that. Oh yeah, X percent of our revenue comes from AI services. Well, what the hell does that mean? I've seen an Amazon job ad historically that listed S3 as an AI service. So if you squint hard enough with AI being shoved into everything, we're not talking apples To Apples here. We're wishcasting.
A
Okay, I'm going to do another Matt Garman thing because that interview was good. If you haven't seen it, GeekWire's got a bunch of excerpts from it. So he said, AWS is terrible at being a fast follower, which caught my eye because I always think about Microsoft as doing best as a fast follower instead of best first to market. Like when they were first to market with the OpenAI stuff, I'm like, oh, this doesn't bode well in my view. I think this is going to be bad. So I'm curious if you guys agree, does it even matter who's the fastest or slowest follower of them all? And where does Amazon and where does Microsoft sit in that world?
B
It matters. This quarter when GPT came out, or GPT as I insist on calling it, was released, yeah, there was nothing quite like it. I spun up an Azure account explicitly to wind up hitting the API. At this point in time, though, let's be realistic here. There is no meaningful difference between the frontier models that I have been able to find. So there's this. I'm in this space now where, great, I'll use whatever is convenient and close and solves my problem the best. If Anthropic goes under, well, Gemini is right there for most use cases. Those have become largely commoditized. So being the first to market with one of these things, okay, maybe it matters to this quarter's numbers, but I don't see that it'll matter a year from now. It's going to be a question of integration of where I'm already in the ecosystem. If everything else I'm building lives on aws, I'm not going to want to go cross cloud to hit some other API providers most of the time.
C
Right? Yeah, you're going to want to go to a power source that's close to your building. And I totally agree with you, Corey. I think that's where we're at right now is that things are commoditized. And I think that's the problem I again see with Microsoft is that they're following well, they're trying to keep up with everybody else. And we were at such a point where everything's pretty much neutralized. And again, because Copilot isn't tied back together with it really well, they're not winning or losing anything. And I think again, Copilot's kind of harming their AI very much, I think
B
that we're going to see a future where the percentage of the AI market share is just a reflection of the rest of their cloud market share. Because this is. Oh, this is where all this is where, I don't know, 60% of infrastructure lives. Great. That's where 60% of AI spend is going to be as well. There are some inflection points because Amazon doesn't have anything that's customer facing in the general sense that's any good. So. Okay, yeah, there might be an Office365 uplift if you can ever get it to stop being annoying and start being useful. But. But that's not going to move the entire universe on its axis, not the way they're talking about it.
A
Right? All right, guys, I'm going to close there and I want to say thank you so much, Corey Quinn for joining us. And Wes, that was a fun conversation. I loved it.
C
Definitely. Thank you for joining, Corey.
B
Yes, thank you for your hospitality and again, indulging my ongoing love affair with the sound of my own voice.
A
All right, I want to remind our listeners they can find lots more coverage of all things Microsoft related on Directions on Microsoft Microsoft.com thank you so much for listening. If you have questions, comments or any topics you would like to hear the directions analysts and guests cover in one of these podcasts, please do not hesitate to contact me via X or Blue Sky. Directions on Microsoft is on LinkedIn, so make sure you give us a follow there and give us a follow at DirectionsMSFT on X or directions on Microsoft on BlueSky for all of the latest Microsoft Enterprise product and licensing information. Thanks again.
Podcast: The Directions on Microsoft Briefing Podcast
Host: Mary Jo Foley
Guests: Corey Quinn (Chief Cloud Economist, Duckbill), Wes Miller (Directions on Microsoft Analyst)
Release Date: December 10, 2025
This episode explores the increasingly commoditized nature of cloud computing in the AI era, specifically the competition between Microsoft Azure and AWS, and the changing dynamics with the entrance of AI platforms and Google Cloud. Mary Jo Foley interrogates whether there is still meaningful differentiation between cloud hyperscalers, or if the cloud has simply become another commodity service, particularly as artificial intelligence rises to the forefront. The discussion also touches on branding challenges, market share shifts, the blurred lines between cloud providers, and the real value of AI and “agent factories” in the current cloud landscape.
Quote (Mary Jo Foley, 00:20):
"So my question is—Is there any competitive advantage for one of these hyperscalers over the other anymore? Or are we in a world where AI has commoditized the cloud?"
"If I had to pick a provider that had differentiated capabilities out of the hyperscalers, I'd say Google these days."
"Everybody sort of evened out at this point. So…we'll see where they go from here."
Corey Quinn: Argues that a small percentage of cloud spend goes into AI, with the majority supporting traditional infrastructure. Azure is "generally less expensive," but not due to advantages—rather, due to historical weaknesses (like security).
"You remove the names...you can't pick up the difference between one to the other in terms of feature approaches, infrastructure reliability."
Advises enterprises to "pick a horse and go all in" rather than chase minute differences: "I'm not a partisan here."
Wes Miller: Still sees the majority of spend as traditional, not AI-driven. Recommends staying with one cloud partner for AI if you’re already invested there.
"I keep hearing Google, you know, at the platform layer. I don't hear Google a lot still...but things like Gemini…are getting more mindshare than Microsoft’s…cavalcade of Copilots."
"Even my customers don't really know how much they're spending on Azure versus the rest of Microsoft…for workloads that are Internet facing, I see GCP more than I see Azure."
"It's not an advantage...there's no virtuous cycle getting organizations to say, you know what, this is built on that great platform in Azure."
"It's effectively what if Microsoft Clippy came back, only now he's been reading your mail."
"Great. AWS is a $132 billion business annually…It’s over 1 to 2% of our annual revenue. Good for you."
"There is no meaningful difference between the frontier models that I have been able to find...these have become largely commoditized."
Mary Jo Foley, 00:20:
"Are we in a world where AI has commoditized the cloud?"
Corey Quinn, 02:57:
"If I had to pick a provider that had differentiated capabilities out of the hyperscalers, I'd say Google these days."
Wes Miller, 07:08:
"I keep hearing Google, you know, at the platform layer...things like Gemini are getting more mind share than Microsoft’s…cavalcade of Copilots."
Corey Quinn, 12:56:
"It's effectively what if Microsoft Clippy came back, only now he's been reading your mail."
Corey Quinn, 18:49:
"There is no meaningful difference between the frontier models that I have been able to find...these have become largely commoditized."
This episode gives listeners a grounded, sometimes irreverent analysis of today's cloud computing landscape—one where the once-fierce rivalries between providers are blurring, AI platforms are mostly commoditized, and true differentiation comes less from technical features and more from strategic integration, branding clarity, and the scale of existing customer lock-in. Despite the marketing noise, real-world enterprise choices remain guided by pragmatic, long-term investments—pick a platform, and don’t expect AI alone to rewrite the rules. Both guests agree: the cloud is already a commodity, and AI is following in its footsteps.